Colloquium Cognitive Systems

Bio-inspired acoustic sensors to improve human and machine hearing

Prof. Dr. Claudia Lenk (Ulm University)

Abstract:  Human hearing has remarkable properties not only in terms of dynamic range and frequency resolution but also regarding speech perception in hard-to-hear, noisy situations. The latter being known as cocktail party effect. Technological speech processing systems exhibit quite remarkable performance as well, in particular, if these include bio-inspired signal pre-processing stage (e.g. nonlinear amplification and frequency filtering) and extensive noise reduction algorithms. However, these systems still struggle with speech processing for changing acoustic environments, in particular, at low signal-to-noise ratios, and if energy and computing power is limited, e.g. in hearing aids.  To tackle these issues, we develop bio-inspired (adaptive, noise-robust) MEMS-based acoustic sensors with integrated processing capabilities, which highlight important features of signals to improve the speech processing performance in noisy environments and the overall efficiency of the sound processing systems for the application in hearing aids, robots/drones and speech processing systems.

About: Claudia Lenk studied technical physics at the TU Ilmenau. She did a PhD in biophysics, working on numerical and chemical modelling of mechanisms underlying atrial fibrillation. During this time she started working in the Electrical Engineering faculty with a focus on scanning probe lithography, micro- and system fabrication and microelectromechanical systems (MEMS). In her Postdoc, Claudia Lenk combined her previous experience by building artificial hair cells using MEMS-devices to improve hearing for humans and machines. Since 2024, she is full professor at Ulm University heading the group biomedical sensor systems and microsystems.